STATS 32: Introduction to R for Undergraduates

Elena Tuzhilina

Sep 21, 2021

http://web.stanford.edu/~elenatuz/courses/stats32-aut2021/

Instructor

Elena Tuzhilina elenatuz@stanford.edu

Born and raised in Moscow, Russia
Major in Mathematics, Moscow State University, 2015
Currently 5th-year PhD student in Statistics
Research interests: applied statistics, especially, with applications in biology
Work experience with ABBYY Lingvo and Microsoft (computer linguistics, data science)

Agenda for today

Course overview: Why learn R?
Class logistics
Installing R, RStudio and packages
Variable types

The big data explosion

John Naisbitt (1982): “We are drowning in information but starved for knowledge.”
We need computing tools to make sense of data

What is R?

S programming language: Developed in the mid-1970s in Bell Labs
- An interactive approach to statistical computing: “… to turn ideas into software, quickly and faithfully”
- Specifically designed for data analysis and graphics
R is a version of S, with some ideas inspired by Scheme
Created by Ross Ihaka and Robert Gentleman at the University of Auckland, New Zealand
Conceptualized in 1992, announced on S-news mailing list in Aug 1993, initial version released in 1995, 1st stable version in 2000
Currently developed by the R Development Core Team

Ross Ihaka & Rob Gentleman

Why learn R?

Reason #1: R was specifically designed for statistics and data analysis.

Many features to aid data analysis which other programming languages don’t have
Interactive, well-suited for exploratory data analysis and rapid prototyping
Good graphical capabilities

Example: Map of 2016 U.S. presidential elections

Example: Spotify Top 100 Songs in 2017

Why learn R?

Reason #2: It’s popular!

(Source: stack overflow)

Why learn R?

Reason #3: It’s easy to get started with R.

Open source: free!
Several online tutorials and resources
A lot of documentation (i.e. how-to pages)
Active community developing “packages”
RStudio
- Open-source integrated development environment (IDE) for R
- Several features which make programming and data analysis easier

Stack Overflow

Q&A site for programmers

Packages

Many people around the world are trying to do the same thing, why not share code?
A package is a collection of R functions, data, documentation and tests
R comes in-built with a core set of packages
- E.g. base, datasets, graphics, stats
Most user-created packages are available (for free) on The Comprehensive R Archive Network (CRAN)
- Packages are generally well-maintained and have good documentation
- Currently, the CRAN package repository features 18,214 available packages.
GitHub is also a popular option for publishing a package nowadays
For packages related to bioinformatics, see Bioconductor

(Source: Revolution Analytics)

Why learn R?

Reason #4: Analyses done in R are reproducible.

R scripts (.R): Lines of R code meant to be run together
R markdown files (.Rmd): Document format that combines code and text into a presentable document

R script

# load packages and get dataset
library(ggplot2)
data(mtcars)

# plot of miles per gallon vs. horsepower, colored by no. of cylinders
ggplot(data = mtcars, aes(x = hp, y = mpg, col = factor(cyl))) +
    geom_point() +
    labs(title = "Miles per gallon vs. horsepower")

R markdown: input

R markdown: output

Why learn R?

Reason #5: Community

Active online community (package development, online discussion and Q&A)
Physical meet-ups
- Full list here
- Bay Area useR Group
- R-Ladies San Francisco
- rstudio::conf
- useR!

How about you?

In the next minute, introduce yourself to someone around you!

Name
One interesting fact about yourself
Why do you want to learn R?

Course objectives

By the end of this course, students will be able to:

Navigate the R ecosystem at a basic level (RStudio, CRAN, R markdown, documentation)
Open, manipulate, plot and save data
Perform simple data modeling and statistical analyses in R

Tentative overview of the course

Week 1: Introduction to RStudio, basic objects in R

Week 2: Data visualization

Week 3: Transforming and cleaning data

Week 4: Importing and publishing data

Week 5: Applications (map-making, data modeling)

Class logistics

10 sessions over 5 weeks (Weeks 1-5)
Tuesday & Thursday 11:30 AM - 1:00 PM, Littlefield 107
Course website: https://web.stanford.edu/~elenatuz/courses/stats32-aut2021/
Office hours: Fridays 9:00 AM - 11:00 PM, Zoom
Recommended text: R for Data Science by Garrett Grolemund and Hadley Wickham

Bring your laptop to class!

Assignments

Programming exercises (ungraded) after each class
Final project proposa (20%) due Oct 20
Final project (80%) due Oct 30

Installing R to your computer

Please follow the instructions on http://web.stanford.edu/~elenatuz/courses/stats32-aut2021/lectures.html

What is a variable?

An “envelope” or “bucket” where information can be stored and referenced
An abstraction that allows us to deal with the underlying structure and relatioships
Each “bucket” has a name on the outside, contains information inside
- Variable name: name of the bucket
- Variable value: information on the inside
Variable assignment: process of putting information inside a bucket

x <- 3

x <- 3

x <- 3
y <- "abc"

x <- 3
y <- "abc"

x <- 3
y <- "abc"
y <- 5

x <- 3
y <- "abc"
y <- 5

x <- 3
y <- "abc"
y <- 5
x <- y

x <- 3
y <- "abc"
y <- 5
x <- y

x + y   # 5 + 5 = 10

Variable types

Every programming language has the notion of data types
The object’s type determines what operations can be done with it
- E.g. You can add 2 numbers, but can you add a number and a letter?
Common types
- integer (…, -1, 0, 1, 2, …)
- double (real numbers),
- character (“a”, “R is cool”, “the quick brown fox…”, “1”)
- boolean (TRUE or FALSE)

Confusion: 123 vs. “123”

How to differentiate between numeric variables and character variables which consist of digits?

Should I treat it as a number like I do in math?
- e.g. Does it make sense to add 2 of them together?
- e.g. Does it make sense to compare their relative sizes?
Numeric variables: counts, age, …
Character variables: Phone numbers, ZIP codes, …

Optional material

List of useful packages

dplyr: Transform data
ggplot2: Make nice plots
readr: Import data into R
tidyr: Clean data
stringr: Tools for working with character strings and regular expressions
lubridate: Make working with dates and times easier
caret: Tools for training regression and classification models
maps, ggmap: Tools for plotting spatial data
shiny: Make interactive web apps

Other fun R stuff

R shiny dashboard for correlations between various stats for NBA 2017-18 season
Making XKCD-like graphs
Animated plots using various R packages
Visualizations projected on the earth as a globe
Visualizing soccer world cup goals

R-bloggers

Blog aggregator of content contributed by bloggers who write about R

The R Journal

Bi-annual open-access journal: Features short to medium length articles covering topics of interest to R users and developers

R-exercises

Website with both tutorials and exercises

DataCamp

Website for learning data science, R included (some courses free, some not)